Learning, Stabilization and Credibility: Optimal Monetary Policy in a Changing Economy

This paper investigates optimal monetary policy in an economy, in which the output-inflation trade off faced by the central bank is influenced by two important forces: (i) the presence of uncertain and possibly changing parameters, and (ii) private sector expectations regarding the central bank's policy. Beliefs regarding the uncertain and possibly time-varying parameters are normal distributions, and are updated according to Bayes rule. Optimal decisions by the central bank involve a certain degree of experimentation. We approximate optimal policies and payoffs using numerical dynamic programming methods and investigate how the incentive for experimentation varies with the extent of parameter uncertainty regarding the short-run slope of the Phillips curve and the weight given to forward-looking private sector expectations in inflation determination. Preliminary findings suggest that the central bank will be willing to repeatedly undertake costly experiments. In other words, the policymaker will tolerate some level of steady-state fluctuations, because they provide information about policy tradeoffs.